-
Notifications
You must be signed in to change notification settings - Fork 1.8k
[TRTLLM-5966][feat] Helix: make softmax stats pointer available to attention gen #6865
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[TRTLLM-5966][feat] Helix: make softmax stats pointer available to attention gen #6865
Conversation
📝 WalkthroughWalkthroughRenamed and unified the softmax statistics parameter to softmax_stats, added it to shared enqueue parameters, and wired it through context and generation paths. Adjusted validation and printing, and updated FMHA/MLA code to consume the new field. Changes
Sequence Diagram(s)sequenceDiagram
participant Caller
participant THOP AttentionOp
participant Common AttentionOp
participant Kernel (FMHA/MLA)
Caller->>THOP AttentionOp: enqueue(..., softmax_stats_tensor?)
THOP AttentionOp->>THOP AttentionOp: validate tensor shape (last dim == 2)
THOP AttentionOp->>Common AttentionOp: common_enqueue_params.softmax_stats = float2*
Common AttentionOp->>Kernel (FMHA/MLA): launch with params.softmax_stats
Kernel (FMHA/MLA)-->>Common AttentionOp: read/write softmax stats
Common AttentionOp-->>Caller: completion
Estimated code review effort🎯 3 (Moderate) | ⏱️ ~20 minutes Suggested reviewers
✨ Finishing Touches
🧪 Generate unit tests
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. 🪧 TipsChatThere are 3 ways to chat with CodeRabbit:
SupportNeed help? Create a ticket on our support page for assistance with any issues or questions. CodeRabbit Commands (Invoked using PR/Issue comments)Type Other keywords and placeholders
Status, Documentation and Community
|
3951b54 to
da4a317
Compare
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Actionable comments posted: 1
🧹 Nitpick comments (2)
cpp/tensorrt_llm/common/attentionOp.h (2)
111-115: Clarify softmax_stats layout and constraints; the current comment is misleadingThe buffer is validated as having a last dim of size 2 in the THOP bridge, but the comment here says shape [num_tokens, num_heads_q] and mentions “LSE/denominator,” which is ambiguous. Please document the expected shape and dtype to prevent misuse.
Apply this diff to tighten the documentation:
- // optional when compute attention stats (MLA chunked prefill or Helix parallelism) - // this is a buffer of size [num_tokens, num_heads_q] with each element - // representing the max and LSE/denominator of the softmax values + // Optional: compute attention stats (MLA chunked prefill or Helix parallelism). + // Device buffer with shape [num_tokens, num_heads_q, 2], where the last dimension packs: + // [0] = max, [1] = log-sum-exp (LSE) of the softmax denominator. + // The tensor must be float32 and contiguous with the last dimension exactly 2. float2* softmax_stats = nullptr;
182-182: Rename print label to match the new field nameThe debug string label still uses the old camelCase name. Update it to keep diagnostics consistent.
Apply this diff:
- ss << "softmaxStatsPtr: " << this->softmax_stats << std::endl; + ss << "softmax_stats: " << this->softmax_stats << std::endl;
📜 Review details
Configuration used: .coderabbit.yaml
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (3)
cpp/tensorrt_llm/common/attentionOp.cpp(2 hunks)cpp/tensorrt_llm/common/attentionOp.h(2 hunks)cpp/tensorrt_llm/thop/attentionOp.cpp(1 hunks)
🧰 Additional context used
📓 Path-based instructions (5)
**/*.{cpp,cxx,cc,cu,h,hpp,hxx,hh,cuh}
📄 CodeRabbit Inference Engine (CODING_GUIDELINES.md)
**/*.{cpp,cxx,cc,cu,h,hpp,hxx,hh,cuh}: In C++, close namespaces with a comment naming the namespace (e.g., } // namespace foo)
Prefer const/constexpr variables over #define for constants
Declare variables const if not modified after initialization
Use Allman brace style in C++
C++ filenames use lowerCamelCase and must be case-insensitively unique within a build target
C++ type names use UpperCamelCase
Local variables, methods, and namespaces use lowerCamelCase
Global non-static variables not in anonymous namespace use gPrefix lowerCamelCase (e.g., gExample)
Static globals or globals in anonymous namespaces use sPrefix lowerCamelCase
Locally visible static variables start with 's' (e.g., static std::once_flag sFlag;)
Member variables use mPrefix lowerCamelCase; public members may omit but are encouraged to use 'm'
Constants (enums, global/static/function-scope magic numbers) use kPREFIXED_UPPER_SNAKE (e.g., kDIGIT_NUM)
If macros are unavoidable, use UPPER_SNAKE_CASE (prefer constants over #define)
Constructor parameter that conflicts with a public member name gets trailing underscore (foo_)
Literal suffixes should be uppercase (e.g., 1234L not 1234l)
C++: use spaces only; indent 4 spaces
Run clang-format (LLVM style) before submitting; wrap lines at 120 characters
If formatting must be bypassed, use // clang-format off/on around the section
Prefer smart pointers; use unique_ptr for sole ownership, shared_ptr for shared; weak_ptr only in exceptional cases
Do not use deprecated pre-C++11 smart pointers
Use C++ style comments; avoid C comments except special inline cases; prefer // single-line
Capitalize and punctuate full-sentence comments
Follow Doxygen rules: use //! for comments and //!< for members in C++
Disable code with #if/#endif and mnemonic conditions; avoid commented-out code; avoid dead code
Do not throw exceptions across library boundaries
Use least-forceful casts; avoid removing const/volatile; avoid C-style and functional casts (except constructors); p...
Files:
cpp/tensorrt_llm/common/attentionOp.hcpp/tensorrt_llm/common/attentionOp.cppcpp/tensorrt_llm/thop/attentionOp.cpp
**/*.{h,hpp,hxx,hh,cuh,cpp,cxx,cc,cu}
📄 CodeRabbit Inference Engine (CODING_GUIDELINES.md)
Parameter names must be consistent between declarations and definitions
Files:
cpp/tensorrt_llm/common/attentionOp.hcpp/tensorrt_llm/common/attentionOp.cppcpp/tensorrt_llm/thop/attentionOp.cpp
**/*.{h,hpp,hxx,hh,cuh}
📄 CodeRabbit Inference Engine (CODING_GUIDELINES.md)
Header files must use include guards named TRTLLM__H without underscores prefix/suffix (e.g., TRTLLM_FOO_BAR_HELLO_H)
Files:
cpp/tensorrt_llm/common/attentionOp.h
**/*.{cpp,cxx,cc,cu,h,hpp,hxx,hh,cuh,py}
📄 CodeRabbit Inference Engine (CODING_GUIDELINES.md)
Prepend NVIDIA copyright header (current year) to all source files
Files:
cpp/tensorrt_llm/common/attentionOp.hcpp/tensorrt_llm/common/attentionOp.cppcpp/tensorrt_llm/thop/attentionOp.cpp
**/*.{cpp,cxx,cc,cu}
📄 CodeRabbit Inference Engine (CODING_GUIDELINES.md)
**/*.{cpp,cxx,cc,cu}: Avoid literal values except for 0, nullptr, true, false; use named constexpr for other literals
Place semicolon of empty for/while loop on a new line
Always use brace-delimited bodies for switch/while/do-for/if/else
Use inline C comments in argument lists when parameter meaning is unclear (e.g., /* checkForErrors = */ false)
Do not use assignment in subexpressions (e.g., if (x = y) ... is forbidden)
Switch on enums should enumerate all values and omit default to catch new values at compile time
Structure switch statements; prohibit fallthrough except between empty cases; each case ends with break or throw; return at end of case not allowed; put break inside braces for compound case
Prefer anonymous namespaces over static for internal linkage of functions
Every defined function must be called at least once (no unused methods)
Files:
cpp/tensorrt_llm/common/attentionOp.cppcpp/tensorrt_llm/thop/attentionOp.cpp
⏰ Context from checks skipped due to timeout of 90000ms. You can increase the timeout in your CodeRabbit configuration to a maximum of 15 minutes (900000ms). (1)
- GitHub Check: Pre-commit Check
🔇 Additional comments (2)
cpp/tensorrt_llm/common/attentionOp.cpp (2)
1773-1774: Good: Context FMHA now receives softmax statsPassing
params.softmax_statsintofmhaParams.softmaxStatsPtrfor context FMHA is consistent with the new field.
1044-1046: FMHA-based runner already wires softmaxStatsPtrThe FMHA‐variant generation path already exposes and sets
softmaxStatsPtr, so parity is ensured. No further changes needed:
- In
fmhaRunnerParams.h,TllmGenFmhaRunnerParamsdefinesfloat2* softmaxStatsPtr;.- In
fmhaDispatcher.cpp(line 203),tllmRunnerParams.softmaxStatsPtr = reinterpret_cast<float2*>(runnerParams.softmax_stats);
|
/bot run |
|
PR_Github #15134 [ run ] triggered by Bot |
|
PR_Github #15134 [ run ] completed with state |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM.
da4a317 to
258082a
Compare
|
@yuxianq @PerkzZheng I believe I addressed your comments, can you quickly review again, please? |
|
/bot run |
|
PR_Github #18812 [ run ] triggered by Bot |
|
PR_Github #18812 [ run ] completed with state |
|
/bot run |
|
PR_Github #18913 [ run ] triggered by Bot |
|
PR_Github #18913 [ run ] completed with state |
258082a to
925c6be
Compare
|
/bot run |
|
PR_Github #18979 [ run ] triggered by Bot |
|
PR_Github #18979 [ run ] completed with state |
|
/bot run |
|
PR_Github #19003 [ run ] triggered by Bot |
Signed-off-by: Matthias Jouanneaux <mjoux@nvidia.com>
Signed-off-by: Matthias Jouanneaux <mjoux@nvidia.com>
Signed-off-by: Matthias Jouanneaux <mjoux@nvidia.com>
b7de48f to
98778f9
Compare
|
/bot run |
|
PR_Github #19012 [ run ] triggered by Bot |
|
PR_Github #19003 [ run ] completed with state |
|
/bot run |
|
PR_Github #19019 [ run ] triggered by Bot |
|
PR_Github #19012 [ run ] completed with state |
|
PR_Github #19019 [ run ] completed with state |
|
/bot run --disable-fail-fast |
|
PR_Github #19036 [ run ] triggered by Bot |
|
Previous unrelated failure waived an hour ago. |
|
/bot run --disable-fail-fast |
|
PR_Github #19040 [ run ] triggered by Bot |
|
PR_Github #19036 [ run ] completed with state |
|
PR_Github #19040 [ run ] completed with state |
…le to attention gen (NVIDIA#6865)" This reverts commit 022d778.
…le to attention gen (NVIDIA#6865)" This reverts commit 022d778. Signed-off-by: Jin Li <59594262+liji-nv@users.noreply.github.com>
…tention gen (NVIDIA#6865) Signed-off-by: Matthias Jouanneaux <mjoux@nvidia.com> Co-authored-by: brb-nv <169953907+brb-nv@users.noreply.github.com>
…tention gen (NVIDIA#6865) Signed-off-by: Matthias Jouanneaux <mjoux@nvidia.com> Co-authored-by: brb-nv <169953907+brb-nv@users.noreply.github.com>
…tention gen (NVIDIA#6865) Signed-off-by: Matthias Jouanneaux <mjoux@nvidia.com> Co-authored-by: brb-nv <169953907+brb-nv@users.noreply.github.com>
Summary by CodeRabbit
New Features
Bug Fixes
Refactor
Description
This PR makes the softmax stas available for tllm-gen attention gen, instead of just context.
Test Coverage
GitHub Bot Help
/bot [-h] ['run', 'kill', 'skip', 'reuse-pipeline'] ...Provide a user friendly way for developers to interact with a Jenkins server.
Run
/bot [-h|--help]to print this help message.See details below for each supported subcommand.
run [--reuse-test (optional)pipeline-id --disable-fail-fast --skip-test --stage-list "A10-PyTorch-1, xxx" --gpu-type "A30, H100_PCIe" --test-backend "pytorch, cpp" --add-multi-gpu-test --only-multi-gpu-test --disable-multi-gpu-test --post-merge --extra-stage "H100_PCIe-TensorRT-Post-Merge-1, xxx" --detailed-log --debug(experimental)]Launch build/test pipelines. All previously running jobs will be killed.
--reuse-test (optional)pipeline-id(OPTIONAL) : Allow the new pipeline to reuse build artifacts and skip successful test stages from a specified pipeline or the last pipeline if no pipeline-id is indicated. If the Git commit ID has changed, this option will be always ignored. The DEFAULT behavior of the bot is to reuse build artifacts and successful test results from the last pipeline.--disable-reuse-test(OPTIONAL) : Explicitly prevent the pipeline from reusing build artifacts and skipping successful test stages from a previous pipeline. Ensure that all builds and tests are run regardless of previous successes.--disable-fail-fast(OPTIONAL) : Disable fail fast on build/tests/infra failures.--skip-test(OPTIONAL) : Skip all test stages, but still run build stages, package stages and sanity check stages. Note: Does NOT update GitHub check status.--stage-list "A10-PyTorch-1, xxx"(OPTIONAL) : Only run the specified test stages. Examples: "A10-PyTorch-1, xxx". Note: Does NOT update GitHub check status.--gpu-type "A30, H100_PCIe"(OPTIONAL) : Only run the test stages on the specified GPU types. Examples: "A30, H100_PCIe". Note: Does NOT update GitHub check status.--test-backend "pytorch, cpp"(OPTIONAL) : Skip test stages which don't match the specified backends. Only support [pytorch, cpp, tensorrt, triton]. Examples: "pytorch, cpp" (does not run test stages with tensorrt or triton backend). Note: Does NOT update GitHub pipeline status.--only-multi-gpu-test(OPTIONAL) : Only run the multi-GPU tests. Note: Does NOT update GitHub check status.--disable-multi-gpu-test(OPTIONAL) : Disable the multi-GPU tests. Note: Does NOT update GitHub check status.--add-multi-gpu-test(OPTIONAL) : Force run the multi-GPU tests in addition to running L0 pre-merge pipeline.--post-merge(OPTIONAL) : Run the L0 post-merge pipeline instead of the ordinary L0 pre-merge pipeline.--extra-stage "H100_PCIe-TensorRT-Post-Merge-1, xxx"(OPTIONAL) : Run the ordinary L0 pre-merge pipeline and specified test stages. Examples: --extra-stage "H100_PCIe-TensorRT-Post-Merge-1, xxx".--detailed-log(OPTIONAL) : Enable flushing out all logs to the Jenkins console. This will significantly increase the log volume and may slow down the job.--debug(OPTIONAL) : Experimental feature. Enable access to the CI container for debugging purpose. Note: Specify exactly one stage in thestage-listparameter to access the appropriate container environment. Note: Does NOT update GitHub check status.For guidance on mapping tests to stage names, see
docs/source/reference/ci-overview.mdand the
scripts/test_to_stage_mapping.pyhelper.kill
killKill all running builds associated with pull request.
skip
skip --comment COMMENTSkip testing for latest commit on pull request.
--comment "Reason for skipping build/test"is required. IMPORTANT NOTE: This is dangerous since lack of user care and validation can cause top of tree to break.reuse-pipeline
reuse-pipelineReuse a previous pipeline to validate current commit. This action will also kill all currently running builds associated with the pull request. IMPORTANT NOTE: This is dangerous since lack of user care and validation can cause top of tree to break.